Particle Value Functions

Abstract

The policy gradients of the expected return objective can react slowly to rare rewards. Yet, in some cases agents may wish to emphasize the low or high returns regardless of their probability. Borrowing from the economics and control literature, we review the risk-sensitive value function that arises from an exponential utility and illustrate its effects on an example. This risk-sensitive value function is not always applicable to reinforcement learning problems, so we introduce the particle value function defined by a particle filter over the distributions of an agent's experience, which bounds the risk-sensitive one. We illustrate the benefit of the policy gradients of this objective in Cliffworld.

Cited by in corpus (5)

• Uncertainty Decomposition in Bayesian Neural Networks with Latent Variables
• Local Search for Policy Iteration in Continuous Control
• Variational Empowerment as Representation Learning for Goal-Based Reinforcement Learning
• Optimized Auxiliary Particle Filters: adapting mixture proposals via convex optimization
• Direct Policy Gradients: Direct Optimization of Policies in Discrete Action Spaces